Experiments show that interference with CaMKI I after LTP can erase LTP, a strong indication of the importance of CaMKll in the LTP maintenance process. CaMKll holoenzymes contain 12 catalytic subunits. In the ON state, each subunit is phosphorylated and therefore active. When a site becomes dephosphorylated, it can be refreshed (rephosphorylated) by a neighboring active subunit. Aim 1. Is refresh dependent on activity? The CaMKll refresh process requires low levels of Ca, but whether achievement of this level is dependent on spontaneous neural activity is not known. Therefore, a fundamental question of interest is whether the maintenance of LTP or memory requires neural activity. In Aim 1A, we will test this in acute hippocampal slices. In Aim 1B, we will explore whether activity is necessary for maintaining memory at the behavioral level using Drosophila. Although it is known that CaMKll is important for Drosophila memory, experiments have not yet tested whether CaMKll is important in memory maintenance. Given the importance of this issue for interpreting the effects of activity on memory, we will conduct the critical erasure test for determining whether CaMKll mediates memory storage in Drosophila. Aim 2. Does CaMKll subunit exchange occur in vivo: a potential mechanism for molecular refresh? According to theoretical models, switch stability could long outlive the lifetime of any subunit if CaMKll underwent protein turnover by subunit exchange: a newly inserted unphosphorylated subunit could be phosphorylated by a neighboring phosphorylated subunit, thereby providing a molecular refresh. We will make the first attempts to test whether subunit exchange occurs in living cells (Drosophila and hippocampus) and characterize its activity-dependence. Aim 3. Computational modelling: what kinds of neural activity are required to refresh CaMKll phosphorylation? The level of resting Ca2+, and that during spontaneous action potentials or mEPSPs can be estimate, as well as the rate of these reactions. We will use a verified computational model of CaMKll to determine whether these brief Ca2+ events are sufficient to refresh the phosphorylated state of CaMKll and thus ensure the stability of stored information. The results will bear importantly on the fundamental question of whether refresh reactions are mediated by spontaneous activity, or alternatively, are dependent on a network process that replays memories. RELEVANCE (See instructions): Understanding the processes that store memory at synapses will have major implications for several health problems. In particular, addiction has been demonstrated to involve persistent changes in CaMKll at synapses in the basal ganglia networks that are critical for addictive behaviors. The proposed work may provide ways to turn off CaMKll and thus reduce addictive behaviors. CaMKll has also been strongly implicated in memory disorders and stroke.